← Every exhibit

Table Table 2 From the research bench

Which mechanisms can create the gap?

MechanismWhat changesSynthetic commercial exampleFirst validation question
Construct underrepresentationThe indicator omits relevant parts of the declared domainLogin count records access but not completion of the promised workflowWhich facets of the construct are absent from the rule?
Construct contaminationThe indicator includes causes or signals outside the targetEmail count includes automation, list size, and administration as well as selling workWhich non-target processes also move the indicator?
Recording or transcription errorThe intended observation is entered, transferred, or coded incorrectlyA stage is saved one step late or a currency code is droppedCan the source event, version, and transformation be replayed?
Temporal or grain mismatchThe indicator and target refer to different units or periodsA current stage is compared with a quarter-end outcome after the stage was editedDo unit key, timestamp, and observation window align?
Threshold misclassificationA continuous or ambiguous condition is forced into a category“Qualified” changes when the reviewer or threshold changesIs the rule versioned and are borderline cases inspectable?
Missingness or non-observationA value is absent, censored, or unavailable rather than zeroNo report export is recorded because the event was outside the instrumentIs the missing state distinct from non-occurrence?

Swipe or scroll horizontally if the table is wider than your screen.

Cite Embed

Reference & Evidence

Source: Table from this essay. Sources and interpretation are given in the article.

Related exhibits